FIG. 1 illustrates a conventional exposure apparatus 100 used for optical lithography. An illumination system 110 (e.g., ArF laser), illuminates (exposes) 120 a pattern 135 in a reticle 130. The patterned light (image light) 140, carrying an image 175 of the pattern 135, is projected 160 via a projection lens 150 onto a wafer 170, which has a photoresist layer (not shown) spin-coated thereon. The image light 160, which can pass through a medium 165 having a particular index of refraction (e.g., water), exposes the photoresist, which is then developed (non exposed photoresist removed for negative-type photoresist and exposed photoresist removed for positive-type photoresist) and etched (e.g., via plasma etching) to form structures in the wafer 170. The accuracy of the transfer of the pattern 135 of the reticle 130 into the wafer 170 depends on several quality issues. For example, aberrations that are residual in the projection lens can result in an image 175 that does not accurately represent the pattern 135 on the reticle. This is one reason that detectors are used to sample the image light 160 incident on a wafer 170.
The resolution R of the conventional system illustrated in FIG. 1 is determined by its projection lens 150 numerical aperture (NA) and the illumination (exposure) wavelength (λ). The relationship can be expressed as:R=(k1λ)/NA  (1)
where k1 is a process dependent factor (e.g., between 0.3-0.5). For example, a NA of over 1.0 can be obtained using an immersion system. ArF laser (A=193 nm) is typically used as the illumination system. For an immersion system the space 165 between the projection lens 150 and the wafer 170 is filled with fluid. The fluid can be transparent to the illumination wavelength and have an index of refraction “n” greater than 1 (e.g., purified water n=1.44). Note that when the term fluid is referred to herein it can include liquids (e.g., water) and gases (e.g., air at various pressures).
One method of conventional detection to acquire the image accuracy is to expose the photoresist, develop the photoresist, and view it under a scanning electron microscope (SEM). However several photoresist properties in addition to the desire to sample the image directly have led to the development of sampling detection arrangements.
FIG. 2A illustrates a conventional one slit (aperture) sampling detection arrangement 200a (detector aperture) used to sample the image light to acquire a measure of the accuracy of the image 210a. The image light is incident on the detector aperture whereby a portion of the light is shielded from passage through the detector aperture by a shield layer 230a. The shield layer 230a can lie on a support substrate 240a, and has an opening (aperture) 220a through which light of a chosen wavelength can pass. The portion of image light 250a, which passes through the aperture, is detected by a detector 260a. To sample the entire image the detector aperture is moved 290 along with the detector 260a. 
When using a single aperture, a non-periodic image (as opposed to the periodic image illustrated in FIG. 2A) can be detected. For example FIG. 2B illustrates a single slit system as illustrated in FIG. 2A, as applied to a non-periodic multiple intensity image 210b. 
In the case of a periodic image in the scan direction a multi slit system can be used. FIG. 2C illustrates a conventional multi slit 220b (multi-aperture) image sampling detection arrangement 200b (detector aperture). This arrangement is similar to the single aperture arrangement but with multiple slits to sample a periodic image. Both arrangements can be moved to sample the entire image. Since the image is periodic, each aperture practically captures the same image portion. The arrangement in FIG. 2C allows more light, compared with the single slit arrangement in FIG. 2A, to be detected by the detector. Again the image 210c is sampled by a detector 260b that detects the portion of the image light 250b passing through the detector aperture 200b, i.e. the portion passing through the multi-apertures 220b. The other portions not being transmitted through the detector aperture 200b are shielded by the shield layer 230b, which is supported by the support substrate 240b. 
In several conventional arrangements the aperture width (e.g., a slit width) is smaller than the image light wavelength so that image features can be detected. U.S. Pat. No. 5,631,731 discusses a single slit aperture arrangement and a multi-slit arrangement. Note that the term aperture is used to denote an opening, a slit, a hole, and any other type of region that allows a particular frequency of light to pass while other neighboring regions do not. Since the aperture width tends to be smaller than the wavelength of the image light, diffraction can occur upon exiting the aperture(s), adding to image detection errors at the detector (e.g., reduced accuracy, reduced intensity). Reduced accuracy of measurement can occur when different polarizations of the image light, each having different levels of contrast, are transmitted differently through the detector aperture. Reduced intensity can occur when only a portion of the diffracted light reaches the detector. Reduced intensity can also be a function of angle of incidence.
FIG. 3A illustrates the diffraction 370 of image light 330, incident to a normal of a surface at an angle theta (θ), passing through a narrow slit aperture 320 of a detection aperture 300. The aperture width is “d” and is assumed to be less than the image light wavelength λ 340. The thickness of the shield layer 310 is W1, and can vary depending upon the extinction characteristics of the material used for the shield layer 310. The image light 330 can have two polarizations, a Transverse Electric (TE) polarization 360a and a Transverse magnetic polarization (TM) 360b. The electric field of the TE mode 360a is substantially aligned with a chosen direction (e.g. along the long part of a slit), while the electric field of the TM mode 360b is aligned substantially perpendicular to the TE mode's field alignment. In a conventional exposure system (FIG. 1) the illumination beam is not polarized. In such a case, the image 175 is given as a superposition of the TE and TM polarization components (roughly 50% each). Thus, if the slit transmittance is different between the TE and TM polarizations, the image detected through the aperture will be different from the image that would have been created in photoresist without the aperture. This can lead to an accuracy problem in the measurement.
As discussed above the different polarizations TM and TE can have different transmittance properties through an aperture. FIG. 3B illustrates the relative transmittance of TE and TM polarized image light through the narrow aperture 320 of the detector aperture 300 of FIG. 3A. The plots are based upon a simulation (solving Maxwell's Equations using Finite Difference Time Domain (FDTD) method) where the conditions assumed are:
that the shielding material is Cr, where the optical properties are n=0.841, and k=1.647;
where the illumination light has a wavelength of λ=193 nm (e.g., ArF laser);
where the thickness of the shielding later is about 95 nm; and
where the aperture is a slit and is varied from 15 nm to 235 nm.
As is illustrated in FIG. 3B the transmittance for the TE and TM modes are different for various slit widths, and match at a slit width of about 60 nm. However to determine features in the image light it is often useful to have a slit width smaller than 60 nmn, for example 45 nm. At 45 nm though the TM transmittance is higher than the TE transmittance. Notice that at 45 nm the transmittance level of both modes have decreased.
To illustrate reduced intensity due to diffractive effects on the transmitted image light, a detector is moved to two positions A1 and A2 (FIG. 3A) and the image light measured and plotted in FIGS. 3C and 3D respectively.
FIG. 3C illustrates the Poynting vector intensity as a function of spatial dimension at an observing position A1 for the simulation conditions. Position A1 is next to the aperture 320 while position A2 is offset 1.5 microns from the aperture exit. FIG. 3C illustrates the localized intensity about the Aperture Middle Line (AML) with no offset distance. However, often the detector can not be placed at the aperture exit. Instead the sensor of the detector is often set within the detector so that there is a resultant offset distance.
FIG. 3D illustrates the Poynting vector intensity as a function of spatial dimension at an observing position A2, at an offset distance of 1.5 microns from the aperture exit. FIG. 3D illustrates the dispersed and reduced nature of the intensity about the AML at an offset distance. Thus the diffractive nature of the aperture can result in decreased intensity at the detector and contrast reduction, and a lower signal to noise ratio (i.e., SN ratio).
Additionally variation in transmitted intensity, through a 1-D slit, between TE and TM can occur as the incident angle θ varies. FIG. 3E illustrates the angular and polarization dependence of transmittance as determined via simulation where (referring to FIG. 3A) the aperture 320 is a cross-section of a 1 D slit filled with SiO2; the aperture width d>=50 nm; the shield layer 310 is made of Tantalum with a thickness W1>=80 nm; the medium 310A is water; and the medium 310B is SiO2. As can be seen the TE and TM polarizations have different transmission properties. With the transmittance of both modes approach each other in value as the incidence angle increases. For accurate measurements, the transmittance between the TE and TM polarizations needs to be matched including their angular behaviors. One solution to this issue has been discussed in application 10025281US01(100-3365) filed on Oct. 21, 2005, which discusses matching the transmittance of the TE and TM polarizations by properly adjusting the effective refractive index in an aperture space. By controlling the volume ratio between mediums (e.g., air and/or other materials) in the aperture (for example, the same material as substrate), the refractive index in the aperture space can be effectively controlled. In the discussed system to avoid intrusion of a liquid medium into the aperture space (e.g., if part of the medium in the aperture is air) a cover plate can be used. Additionally, with the discussed system, the fabrication method can be difficult to obtain a partially filled aperture.
In addition to diffraction's reduction of the detected image intensity, the various transmittances of the TE and TM mode can result in reduced accuracy in the measurement. For example FIG. 4A illustrates the image 420a from TE polarized image light 410a. 
FIG. 4B illustrates the image 420b from TM polarized image light 410b. When the illumination beam is not polarized, the actual image is represented as a superposition (or as the average) of the two image components. Thus, when more TM mode light is transmitted through an aperture the detected image will be different from the image created on the wafer. Thus, conventional detector apertures can suffer from various levels of diffractive effects and accuracy issues.